Effect of gas adsorption on the electrical properties of single-walled carbon nanotubes mats

Abstract

Single wall nanotubes have been made by arc-discharge method. Residual impurities (fullerenes, amorphous carbon, catalyst metals...) have been removed by tangential filtration process followed by high temperature annealing under vacuum (1200 °C). In this work we present results on the influence of the surrounding gas nature (N 2 , H 2 , CO 2 , H 2 O…) on the electrical resistivity of carefully outgassed mat of such samples. In particular, we have observed that the sample resistance exhibits a strong dependence on water contamination during the transfer to the measurement reactor.

Full text

Page 1

1
EFFECT OF GAS ADSORPTION ON THE ELECTRICAL PROPERTIES OF SINGLE
WALLED CARBON NANOTUBES MATS

C. MARLIERE*, P. PONCHARAL**, L. VACCARINI**, A. ZAHAB**
*Laboratoire des Verres, Univ. Montpellier 2, CC 69, 34095 Montpellier, France,
marliere@univ-montp2.fr
** Groupe de Dynamique des Phases Condensées, CC 26, 34095 Montpellier, France.

ABSTRACT

Single wall nanotubes have been made by arc-discharge method. Residual impurities
(fullerenes, amorphous carbon, catalyst metals...) have been removed by tangential filtration
process followed by high temperature annealing under vacuum (1200 °C). In this work we
present results on the influence of the surrounding gas nature (N2, H2, CO2, H2O…) on the
electrical resistivity of carefully outgassed mat of such samples.
In particular, we have observed that the sample resistance exhibits a strong dependence on
water contamination during the transfer to the measurement reactor.


INTRODUCTION

Since their discovery in 1991 [1], carbon nanotubes have been the subject of intensive
research, motivated by the intrinsic richness and diverse applications potential. Their
conductive properties depend drastically on both the diameter and chirality of the hexagonal
carbon lattice along tubes. In fact, a slight change in the winding of hexagons along the tube
can transform the electronic properties of the tube from metallic to a large gap semiconductor
[2].
Several groups have reported electrical resistivity measurement results for mats of single-
wall carbon nanotube (SWNT). Around the ambient temperature, this conductivity exhibits
metallic (i.e., positive dR/dT) or non-metallic (i.e., negative dR/dT) behavior depending on the
experimental preparation (arc discharge or laser ablation) [3-5] and measurements conditions.
Due to their high porosity, these materials are extremely sensitive to doping and/or gas
adsorption. In this paper, we present results on the influence of the surrounding gas nature on
the electrical resistivity of carefully outgassed mats of such samples.

EXPERIMENTAL SET-UP

Four-probe electrical resistance (E.R.) measurements were performed in a high vacuum
chamber on SWNT mats. These samples were produced by the electric arc method [6]. A
graphite rod containing powder of catalyst particles (Ni:Y:C) was vaporized under a He
atmosphere at 660 mbar. Collecting the collaret part, we performed purification by successively
using the following processes: acid treatment, cross flow filtration and annealing under
nitrogen atmosphere up to 1200°C during 6 hours [7]. The final product is a compacted mat
which contains a multitude of cleaned and compacted SWNT bundles with only a small density
of metallic carbides or oxides.
The 0.2x2x25mm3 samples were cut and placed in the E.R. measurement chamber which
was then carefully outgassed (base pressure < 3.10-5 mbar). This chamber was divided in two
compartments connected to the pumping system: the first one contains the sample holder and
its heater; the second one is isolated from the other parts of the chamber allowing the
Oral presentation at MRS 1999 Fall Meeting, Boston
Published in: Material Research Society Symposium Proceedings , 593 (2000) 173

Page 2

2
preparation and thermalization of the high purity (99.995%) gases before introduction –at a
final pressure of ~1,000mbar- in the measurement chamber (volume ~1dm3).
E.R. measurements were done at 25°C by using a 10mA DC current. Parasite
thermoelectric effects were compensated. The sample temperature was measured by a platinum
resistor placed very near from the carbon nanotube (CNT) mat.
Introduction or pumping of gases were done rapidly : the pressure in the reactor was
varying between the working pressure (P) of ~1,000mbar and 1mbar in less than one second.
We also studied the influence of minute quantities of H2O vapor by injecting few microliters of
water of ultra-high purity in the intermediate chamber. Nitrogen was then injected in the reactor
until the total and final pressure reached 1,000mbar. Before each refilling of the reactor with a
new gas a careful outgasing was performed by heating the sample up to 200°C at a constant
rate of about 3°C/mn. The sample was held at this temperature during 5 hours.

RESULTS AND DISCUSSION

Preliminary it must be noted that a pure ohmic behavior was always observed in our
experiments whatever the pressure in the reactor and the nature of the surrounding atmosphere.

A first and simple experiment showed the crucial importance of the surrounding
atmosphere on the electrical properties of the CNT mats. The E.R. variations of samples
previously stored in air were studied during the first thermal outgasing cycle (P<4.10-5mbar):
Figure 1 (dashed line) revealed a highly irreversible behavior. On the opposite the E.R.
variations measured during the next heat-up and cool-down cycles (Fig.1 – full line) were
reversible and identical. Adsorbed gases were desorbed during the first cycle leading to a now
reversible thermal behavior. Thermal coefficient (β) was calculated from the slope of the curve
E.R. versus temperature in the vicinity of the temperature of the experiments (25°C). The value
of β is always negative and has a constant absolute value of about 10-3 °C-1 whatever the
pressure or the gas nature.
The following point has to be emphasized. During all the experiments described below we
verified that the E.R. variation due to temperature variation of the substrate (lower than 0.1°C
during a gas filling and pumping cycle) was negligible when compared to these observed
during the gas ad- or de-sorption experiments.

First we studied the adsorption of monoatomic or symmetrical diatomic gases: N2, H2. O2,
Ar. In all these cases we noted a small but detectable increase of E.R. (~4.10-2 percent in
relative values) after the introduction of these gases. As shown in Figure 2 for N2 this effect is
mainly reversible. However the amplitude of these E.R. variations is slightly decreasing if there
is no careful and efficient outgasing between the different gas introductions. This saturation-
like behavior is probably due to the non-complete outgasing and refilling of the sample.
This E.R. increase (named effect #1 in the following) could be attributed to the insertion of
gas molecules between the interlaced ropes or even between the individual tubes inside the
ropes.

Similar experiments were done with high purity carbon dioxide. Two phenomena were
observed after the gas introduction. In a first step (Fig. 3), a positive variation of E.R. occurred.
Its intensity and time constant are of the same order of magnitude as with nitrogen. The same
explanation as before can reasonably be proposed to interpret this effect.

Page 3

3
In a second step we note a large (one order of magnitude higher than for the effect #1)
exponential-like decrease of E.R. with a time constant of about 20mn. This slow process –
when compared to that involved in effect#1 – could be related to the adsorption of CO2
molecules on the CNT surface acting as an apparent electron donor by means of π-type C=O
bindings. The dissolution of CO2 in residual traces of adsorbed H2O molecules followed by
proton production (acid-base reaction) leading to an increase in conductivity is unlikely to
happen due to the careful outgasing of the sample.

We also studied the water vapor adsorption effects on CNT resistivity. Minute H2O quantity
(the equivalent of 15µlL of liquid water) diluted in pure nitrogen gas (total pressure of 1atm)
was introduced. In a first step (Fig. 4), a small and positive E.R. variation (~10-1 %) likely due
to the effect #1 is observed. Then a very important decrease (around 10%) is observed (with a
time constant of 6mn). This second effect is similar to that observed with CO2 but with much
higher amplitude: H2O molecules are likely adsorbed on or in CNT resulting in an apparent
electron donor behavior. According to Arai et al. [8] a non-dissociative and dissociative
adsorption reactions according to the following scheme can occur:
H2O(g) → H2O+ (ad) + e-
H2O(g) + Oo2- + Vo- → 2OHo- (ad) + e-
H2O(g) + Oo2- + Vo2- → 2OHo- (ad) + 2 e-
Where Vo- and Vo2- are oxygen vacancies trapping one or two electrons.

CONCLUSION
Due to their high porosity, SWNT mats are extremely sensitive to doping and/or gas
adsorption. On these samples electrical resistance was shown to be very dependent on the
nature of the surrounding atmosphere. Two effects giving opposite variations of E.R. are
competing. First a steric effect -observed for all the studied gases- causes a small increase of
E.R. Secondly adsorption of highly polarizable molecules (H2O and CO2) increases notably the
electrical conductivity of these samples. This last effect is likely due to apparent electron donor
behavior.
These observed effects are of crucial importance for interpretation of air-performed
measurements on CNT (isolated, in bundles or in mats): E.R., Raman, X-rays …
According to the experiments CNT revealed themselves as good candidates for sensitive
gas sensors (vapor water, carbon dioxide).

REFERENCES

1. S. Iijima, Nature 354, p. 56 (1991).
2. J.W. Mintmir et al. Phys. Rev. Lett. 68, p. 579 (1992).
3. J.E. Fischer et al. Phys. Rev. B55, p. R4,921 (1997).
4. L. Grigorian et al Phys. Rev. B58, p. R4,195 (1998).
5. L. Grigorian et al. Phys .Rev. Lett. 80, p. 5,560 (1998).
6. C. Journet et al., Nature 388, p. 756 (1997).
7. L. Vaccarini et al., Synthetic Metals 103, p. 2,492 (1999).
8. H. Arai and T. Seiyama, in Sensors : A comprehensive survey, vol. 3, edited by W. Göpel, J.
Hesse, and J.M. Zemel (VCH Verlag, Weinheim, 1992) p.371

Page 4

4

050
TEMPERATURE (°C)
100 150200
3,8
3,9
4,0
4,1
4,2
4,3
Figure 1 : Electrical resistance of SWNT mat (stored in air) during thermal cycles under
vacuum versus temperature. The first cycle is in dashed line the second one in full line.
R = Ro [ 1 + β T(°C) ]
β = - 1.0 10
-3 °C
-1
ELECTRICAL RESISTANCE (Ohms)
0 10 20 30
-0,02
0,00
0,02
0,04
0,06
0,08
0,10
Figure 2 : Electrical resistance relative variations of the SWNT mat during several filling and
pumping cycles with nitrogen versus time.
τ (N2) = 40 s
N2
N2
∆ R / R ( %)
N2
TIME (mn)

Page 5

5

0 102030 405060
-12
-10
-8
-6
-4
-2
0
2
Figure 4 : SWNT mat resistance relative variations versus time during exposure to water vapor
(the first step corresponds to 15 µlL , the others to 5 µlL ). The arrows indicate the moment
of gas filling.
τ = 6.4 mn
∆ R / R ( %)
H2O
TIME (mn)
0 10203040
-0,10
-0,05
0,00
0,05
0,10
Figure 3 : Electrical resistance relative variations versus time during two filling and pumping
cycles with carbon dioxide. The arrows in full (resp. dashed ) line correspond to the filling
(resp. pumping ) of the reactor.
∆ R / R ( %)
CO2
τ = 20 mn
TIME (mn)